Sentence extraction system, sentence extraction method and information storage medium
At least one processor of a sentence extraction system specifies at least one candidate sentence based on a predetermined extraction condition, the candidate sentence being a candidate for a sentence to be extracted from among a plurality of sentences; specifies at least one linked sentence that has a link to the at least one candidate sentence; and determines whether to extract the at least one linked sentence, based on whether to extract the at least one candidate sentence.
1. A sentence extraction system comprising at least one processor configured to:
obtain a score based on an extraction condition for a plurality of candidate sentences,
sequentially select a predetermined number of candidate sentences, beginning with a candidate sentence with a higher score,
select at least one linked sentence for every candidate sentence selected,
wherein the at least one linked sentence has a score that fulfills a predetermined selection criterion,
wherein the at least one linked sentence does not fulfill the extraction condition, and
wherein the at least one linked sentence has a link to the candidate sentence, and
determine whether to extract the at least one linked sentence, based on whether to extract the at least one candidate sentence.
2. The sentence extraction system according to claim 1 , wherein the at least one processor:
obtains a score based on the extraction condition for every candidate sentence, and
changes the at least one candidate sentence whose score fulfills a predetermined change criterion such that the at least one candidate sentence is not extracted.
3. The sentence extraction system according to claim 2 , wherein the at least one processor:
specifies at least one other candidate sentence that has a link to the at least one candidate sentence whose score fulfills the predetermined change criterion, as the at least one linked sentence, and
determines not to extract the at least one linked sentence.
4. The sentence extraction system according to claim 2 , wherein
the change criterion is an order of the score, and
the at least one processor sequentially selects a predetermined number of candidate sentences, beginning with a candidate sentence with a lower score, and changes the predetermined number of candidate sentences selected such that the predetermined number of candidate sentences selected are not extracted.
5. The sentence extraction system according to claim 3 , wherein the score of at least one non-candidate sentence that does not fulfill the extraction condition among the plurality of sentences is set to a predetermined value or is not calculated.
6. The sentence extraction system according to claim 3 , wherein the at least one processor restrains the at least one candidate sentence whose score fulfills a higher criterion than the change criterion from being determined as a sentence not to be extracted even though the at least one candidate sentence has a link to the at least one candidate sentence that fulfills the change criterion.
7. The sentence extraction system according to claim 1 , wherein
a document containing the plurality of sentences is divided into a plurality of parts, and
the at least one processor specifies the at least one linked sentence for every part.
8. The sentence extraction system according to claim 1 , wherein the at least one processor specifies the at least one linked sentence, based on at least one of connection property, referential property, and vocabular link property with respect to the at least one candidate sentence.
9. The sentence extraction system according to claim 1 , wherein
the extraction condition indicates that at least one keyword among a plurality of keywords is contained, and
the at least one processor specifies the at least one candidate sentence that contains the at least one keyword.
10. The sentence extraction system according to claim 1 , wherein
each of the plurality of sentences is related to content of a financial settlement of a company, and
the extraction condition is a condition for extracting a performance factor sentence that describes a factor for performance.
11. A method for extracting a sentence, comprising:
obtaining a score based on an extraction condition for a plurality of candidate sentences,
sequentially selecting a predetermined number of candidate sentences, beginning with a candidate sentence with a higher score,
selecting at least one linked sentence for every candidate sentence selected,
wherein the at least one linked sentence has a score that fulfills a predetermined selection criterion,
wherein the at least one linked sentence does not fulfill the extraction condition, and
wherein the at least one linked sentence has a link to the candidate sentence, and
determining whether to extract the at least one linked sentence, based on whether to extract the at least one candidate sentence.
12. A non-transitory information storage medium storing a program for causing a computer to:
obtain a score based on the extraction condition for a plurality of candidate sentences,
sequentially select a predetermined number of candidate sentences, beginning with a candidate sentence with a higher score,
select at least one linked sentence for every candidate sentence selected,
wherein the at least one linked sentence has a score that fulfills a predetermined selection criterion,
wherein the at least one linked sentence does not fulfill the extraction condition, and
wherein the at least one linked sentence has a link to the candidate sentence, and
determine whether to extract the at least one linked sentence, based on whether to extract the at least one candidate sentence.
13. The sentence extraction system according to claim 1 , wherein the predetermined extraction condition is based on the presence of a keyword.